Gatlin v. St. Louis / AWS Rekognition false arrest suit (2026)
Gatlin v. St. Louis — Amazon Web Services Added as a Defendant in a Wrongful Arrest Suit Where a Facial Recognition Match on a Poor-Quality Image of a MASKED Suspect Put a Man in Jail for 17 Months (Amended Complaint, August 2026)
United States · Gatlin v. City of St. Louis et al. — wrongful arrest action; amended complaint adding Amazon Web Services (August 2026)
- Status
- Blocked / in litigation
- Applies to
- The St. Louis Police Department and St. Louis County police as defendants; AMAZON WEB SERVICES, added as a defendant in August 2026
- Enforced by
- U.S. federal court (U.S. Magistrate Judge; specific district not identified in the located source)
Christopher Gatlin, a St. Louis resident, spent SEVENTEEN MONTHS IN JAIL for an assault he did not commit. The sequence that put him there is the reason this case matters. Police investigating a December 2020 assault on a MetroLink security guard submitted a POOR-QUALITY IMAGE OF A MASKED SUSPECT to the St. Louis Mugshot Recognition Technology system. The system returned Gatlin as a possible match. He had not previously been a suspect — there was no other reason to look at him. On the strength of that match police put his photo in a lineup, where suggestive procedures influenced the victim's identification. He was arrested on August 14, 2021 and could not post a $75,000 bond. The criminal charges were dismissed in March 2024 after body camera footage revealed the improper lineup procedures. Two things make the civil case significant. In October 2025 a U.S. Magistrate Judge denied motions to dismiss most of Gatlin's claims, so the case is proceeding rather than being disposed of early. And in AUGUST 2026 an amended complaint ADDED AMAZON WEB SERVICES AS A DEFENDANT. Facial recognition wrongful-arrest suits have overwhelmingly been brought against police departments; naming the technology vendor puts the question of whether a provider bears responsibility for matches its system returns on poor-quality inputs directly before a federal court. No ruling on the merits has issued.
Who is protected: People subjected to facial recognition searches by law enforcement, particularly where a match generates rather than corroborates suspicion
Legal detail
U.S. federal court — CHRISTOPHER GATLIN's civil action arising from his wrongful arrest and 17-month pretrial detention in St. Louis, Missouri. FACTUAL SEQUENCE: police investigating a DECEMBER 2020 assault on a MetroLink security guard submitted a POOR-QUALITY IMAGE OF A MASKED SUSPECT to the ST. LOUIS MUGSHOT RECOGNITION TECHNOLOGY system; the system returned GATLIN AS A POSSIBLE MATCH; he had been under NO PRIOR SUSPICION. The match caused police to include his photograph in a lineup, in which suggestive procedures influenced the victim's identification. ARREST: August 14, 2021. DETENTION: 17 months, unable to post a $75,000 bond. DISMISSAL: criminal charges dismissed MARCH 2024 after BODY CAMERA FOOTAGE revealed improper lineup procedures. THE TWO FACTS THAT MAKE THIS CASE MATTER: (1) OCTOBER 2025 \u2014 a U.S. Magistrate Judge DENIED MOTIONS TO DISMISS MOST CLAIMS, so the action survived the stage at which facial recognition wrongful-arrest suits are commonly disposed of; (2) AUGUST 2026 \u2014 an AMENDED COMPLAINT ADDED AMAZON WEB SERVICES AS A DEFENDANT. VENDOR LIABILITY IS THE NOVEL QUESTION: the wrongful-arrest cases that have shaped this area were brought against police departments and municipalities, on theories of investigative negligence and suggestive identification procedure. Naming the technology provider asks whether a vendor bears responsibility for the matches its system returns \u2014 particularly on an input its own documentation would characterise as unsuitable, a poor-quality image of a MASKED face. A masked subject is the clearest possible case of an input outside a facial recognition system's reliable operating envelope, and the system returned a name anyway. WHY THE 'NO PRIOR SUSPICION' FACT IS DECISIVE: where facial recognition merely corroborates an existing investigative lead, its contribution to the outcome is arguable; here it GENERATED the lead, so the entire chain of detention traces to the match. RELATED INDEXED MATTERS: ACLU v. Clearview AI and In re Clearview AI Class Settlement concern a different vendor and different theories. STATUS: no merits ruling; strength `unknown`.
Key facts
| Jurisdiction | United States |
|---|---|
| Level | Federal |
| Citation | Gatlin v. City of St. Louis et al. — wrongful arrest action; amended complaint adding Amazon Web Services (August 2026) |
| Private right of action | Yes, individuals can sue |
| Penalties | Damages sought; no judgment has issued |
| Topics | facial recognition · biometric data · police and surveillance AI · consumer protection · automated decision-making |
| Legal analysis | Amazon Facial-ID Tech Targeted in St. Louis False Arrest Suit — Bloomberg Law |
More AI rules in United States
- Trump AI Innovation & Security EO (June 2026) · In effect
- FERC order directing PJM and other grid operators to reform tariffs on… · Enacted (not yet in effect)
- Chatrie v. United States (Fourth Amendment protection for location… · Blocked / in litigation
- Kids Internet and Digital Safety Act (KIDS Act) (H.R. 7757) · Proposed / pending
- NAIC AI Evaluation Tool Pilot (12 states, 2026) · Proposed / pending
- FTC Proposed Policy Statement — AI Accuracy (July 2026) · Proposed / pending
Related facial recognition rules elsewhere
- Virginia § 15.2-1723.2 (facial recognition by local law enforcement, eff. July 1, 2026) · In effect
- Bristol VA PD facial recognition program (first VA local agency) · In effect
- Dallas PD Clearview AI facial recognition policy · In effect
- Ban biometric recognition in public accommodations (Ban The Scan) · Proposed / pending
- Limit facial recognition in residential buildings (Ban The Scan) · Proposed / pending
- Texas v. Google · In effect
See something wrong or out of date? Submit a correction. Every entry must carry a verifiable source.